LOCATION: ShenZhen
TEAM: Yu Liu, Ximing Zhong, Que Guan, Jiaxin Tao
CATEGORY: Exhibition

Mechanical AI can perform hundreds of different scenario designs. Human design perception can be wrapped in a machine. Design practice can become modular and accessible through tools. The line between instrumental and architectural knowledge is becoming increasingly blurred.

Tool-encoded knowledge actually represents a new collective cultural memory.

——MEMEX 1945 

机械AI可以进行数百种不同的方案设计。人类设计感知可以被包裹在机器中。设计实践可以通过工具变得模块化和可获得。工具性知识和建筑知识之间的界限越来越模糊了。

工具编码的知识实际上代表了一种新的集体文化记忆。

——MEMEX 1945 

PART I

Conceptual introduction and theoretical framework

 As the change and deterioration of the global climate and environment, more and more people are concerning about the current human condition, and constantly proposing various sustainable response strategies to solve existing problems and future situations, to seek a balanced and optimized solution strategy. From the point of view of artificial intelligence, we attempt to explore experiments in which artificial intelligence imitates, learns, optimizes, and rapidly generates solutions. Through testing and predicting the climatic environment of the site and learning from high-quality examples, we are able to set different environmental and climatic conditions and needs, and quickly generate sustainable solutions for different geographical environments, site characteristics, climatic types, the relationship between architecture and landscape, and future extensible states. 

During the cloning process, whether the Dolly sheep of architectural design could preserve human intelligence? Could we integrate a cloning and adaptive framework to migrate the order of human decision-making to meet new design solutions and new contexts, where there is decision making, there is the hustle and bustle, and how could we avoid it? Who is more suitable for cloning the decision-making part of human intelligence, machine learning or probabilistic mathematical methods, and we try to explore the deep mathematical and probabilistic logic behind the form, and for machine learning, what theoretical foundations we need to generate valid parts. Through the massive and rapid self-learning, solution optimization, and complex calculations of artificial intelligence, it is hoped that more effective outcome prediction and faster and more rational sustainable solutions will be made in urban new and renovation environments.

随着全球气候、环境的变化与恶化,越来越多人开始关注人类的生存条件和现状,不断提出各种可持续、可平衡的应对策略,以解决现存问题和未来形势,寻求一个平衡且优化的解决策略。我们从人工智能的角度出发,试图去探索人工智能进行模仿、学习、优化并快速的生成解决方案的实验,通过对场地气候环境的测试、预判以及优秀案例的学习,设定不同的环境气候条件与需求,能够快速的针对不同地域环境、场地特征、气候类型、建筑与景观的关系以及未来可延展的状态快速的生成符合条件的可持续景观气候环境。

在建筑设计的多莉羊会在克隆方案的过程中,能保存人类智能决策么?我们是否能够整合一个克隆和适应性框架来迁移人类决策的秩序来满足新的设计方案和新的语境,哪里有决策,哪里就有噪声,我们又该如何避免呢?机器学习和概率数学方法谁更适合克隆人类智能决策部分,我们一起探讨形式背后的深层数学和概率逻辑,以及对于机器学习,我们需要怎样的理论基础来生成有效的部分。通过人工智能大量快速的自我学习、方案优化和复杂计算的进化行为。希望在今后的城市新建环境和改建环境之中做更有效的结果预判以及更快速、更合理的可持续方案。

Material Dimensions and Construction Costs

The whole installation is divided into a central tower and standard units around it, with a height of 2.5m and a length and width of 1m. The materials used include: steel/wood structure, density board, painted aluminium tubes, wire, light strips, spotlights, iron display stands, etc. The overall cost estimate (materials + labour) is around RMB 50,000.

整个装置分为了中心高塔和四周标准单元,高塔高度为2.5米,长宽均为1米,材质包括:钢结构/木结构、密度板、喷漆铝管、铁丝、灯带、、射灯、铁展架等。总体造价预估(材料+人工)5万人民币左右。

Part II

Analysis Chart

Artificial intelligence defines the semantics by learning landscape cases under different climatic conditions, and generates the learned semantics appropriately in the new site context through the given site context.

AI通过学习不同的气候条件下的景观案例,定义语义,通过给定的场地环境,将学习的语义恰当的生成在新的场地环境之中。

We visualised the design and thinking logic of the AI in a 4m x 4m site, cloning and evolving the abstract ai to communicate it physically in the installation, which is divided into a central tower and surrounding standard units, with the surrounding units being cases sampled by the AI, which are filtered by wind and sunlight simulations. The AI learns from the surrounding landscape design solutions and integrates them into the tower as input. These input scenarios are integrated into a single reference scenario.The AI processing unit from bottom to top is shown learning cases, design language definition, design element migration evolution, design element reorganisation and new solution generation. From top to bottom the nodes and the complex computational processes of AI machine thinking are shown. The different colours in the installation represent different design semantics. This is expressed in different colours by way of copper pipes and layered so that the form presented in the lower half of the tower can be seen, while in the centre of the tower the AI performs intelligent solution generation based on the layout and needs of the new site. The spatial semantics that the AI learns are migrated and evolved in the new site, so here we connect and position the same spatial semantics with coloured copper wire, and a new spatial result is generated, with the final new solution presented at the top.

我们在4米×4米的场地中,我们可视化了AI的设计和思考逻辑,把抽象的ai 克隆和进化的用装置实体传达,整个装置分为了中心的高塔和周围的标准单元,周围单元为AI采样的案例,这些案例经过风模拟和阳光模拟分析筛选。 AI从周围的景观设计方案中学习整合到塔中作为输入。这些输入方案被整合成一张参考方案。AI处理装置自下而上分别是学习案例,设计语言定义,设计元素迁移进化,设计元素重组,新方案生成。从上到下展示了AI机器思考的节点和复杂的计算过程。 装置中不同的颜色代表不同的设计语意。 通过铜管的方式用不同颜色表达出来,并将其分层,因此可以看到高塔下半部分呈现的形式,而在高塔的中央,AI根据新场地的布局和需求,进行了智能化方案生成。 AI学习的空间语义在新场地中做了迁移和进化,因此这里我们用彩色铜丝将同一空间语义进行连接定位,便产生了新的空间结果,最终新方案在顶部呈现出来。

Leave a Comment

邮箱地址不会被公开。 必填项已用*标注